[parisc-linux] [PATCH-2.5] Various cleanups

[Joel Soete]

>> how does that work? __c_f_f_c is emitted by the compiler internally,
>> right? how can we define an inline version of this function?
>
>Dooh, you are correct.  I wrote a little test program and the libcall
>doesn't get inlined.
>
>So if no canonicalization is needed, you want something like:
>
>typedef int (*fptr_t) (void);
>
>unsigned int __canonicalize_funcptr_for_compare (fptr_t)
>      __attribute__ ((visibility ("hidden")));
>
>unsigned int
>__canonicalize_funcptr_for_compare (fptr)
>   fptr_t fptr;
>{
>  return (unsigned int)fptr;
>}
>
>The libgcc version is hidden.  That's because shared libraries need
>their own copy which accesses the got table in the shared library.
>Probably in the kernel implementation, a single copy of the routine
>is sufficient and you don't want the hidden attibute.
>
Hi all,

I just try to build a kernel with following stuff:
add first include/asm-parisc/fptr.h:

typedef int (*fptr_t) (void);

extern unsigned int __canonicalize_funcptr_for_compare (fptr_t)
      __attribute__ ((visibility ("hidden")));


and arch/parisc/lib/fptr.c 


#include <asm/fptr.h>

unsigned int
__canonicalize_funcptr_for_compare (fptr)
     fptr_t fptr;
{
    return (unsigned int) fptr;
}

also arch/parisc/kernel/parisc_ksyms.c:
[...]
#include <asm/fptr.h>  
EXPORT_SYMBOL(__canonicalize_funcptr_for_compare);


to finaly arch/parisc/lib/Makefile:
-obj-y   := lusercopy.o bitops.o checksum.o io.o memset.o
+obj-y   := lusercopy.o bitops.o checksum.o io.o memset.o fptr.o

The compile seems ok but knowing not yet anything about modules I would appreciate
your advises before trying to boot it.

Thanks,
    Joel


---------------------------------
Vous surfez avec une ligne classique ?
Economisez jusqu'à 25% avec Tiscali Complete !
Offre spéciale : première année d'abonnement offerte.
... Plus d'info sur http://complete.tiscali.be